Thursday, February 16, 2006

What to Do About Earned Run Averages, Part 3

What to Do About Earneed Run Averages, Part 3
By Carlos Bauer

That’s Nice— But Prove it Works

Next I decided to check it against some pitchers for whom we know how many earned runs they gave up.

First I calculated Joe Pate.  In 1922 Pate gave up 91 ER on way to 24-11 2.71 season for Ft. Worth.  My calculations, using Ft. Worth’s .960 FAVG, give him the same 91 ERs, and the same 2.71 ERA that he had in real life.

Next, Frank Shellenback in 1931.  I give him a 2.67 ERA; his actual was 2.85.

Finally, Fay Thomas in 1934.  2.48 from me against a 2.59 actual ERA.

(What I’m actually doing is calculating an ERA from a mean ER%, and this is what might be characterized as a “Fair Earned Run Average,” or an ERA with all the good and bad breaks taken out.  I use the term “fair” in the same way that fair value is used in the options market. In other words, this tool that was conceived to merely plug in blanks in the records of old-time minor leaguers can be used to check the quality of a current pitcher’s ERA.  For instance, you might suspect a pitcher’s ERA is way too low because he got killed a couple of times, but only a couple of the runs he gave up were earned.)

Next I decided to check my work against some pitchers with great seasons (for all major leaguers, I used The Bill James Electronic Encyclopedia for statistics, so there might be some differences with Total Baseball or The Baseball Encyclopedia:

  1. Bob Gibson in 1968: 1.12 actual, 1.29 estimated.  Great season, but credited with 6 less ERs than would be on average.

  2. Sandy Koufax in 1963:  1.88 actual, 1.69 estimated.  Hard to believe, but his ERA should have been even better than it was.

  3. Bob Welch in 1990: 2.95 actual, 3.13 estimated.  Just about in line.

  4. Lefty Grove in 1931: 2.06 actual, 2.33 estimated.  
All in all, not too bad; all of them very close.  Next I decided to check against some pitchers with horrible records.
  1. Vic Willis in 1905: 3.21 actual, 3.25 estimated.

  2. Paul Derringer in 1933: 3.23 actual, 3.55 estimated.  Figures!

  3. Ben Cantrell in 1935: 4.61 actual, 4.14 estimated.
Next I checked it against the rotation of a team with a low FAVG, in this case Philadelphia NL team in ‘04 (italics are the estimated ERA):
  1. Bill Duggleby, 3.75 and 3.75.

  2. Chick Fraser, 4.50 and 3.92.

  3. Tully Sparks, 2.72 and 2.69.

  4. Fred Mitchell, 4.48 and 4.18.

Finally, I decided to check it against Dick Ellsworth’s career from 1961 through 1968.  (Why I did this only a Cub fan would understand, so I won’t offer any explanation here.)

1961:  3.85 actual, 3.60 estimated.
1962:  5.08 actual, 5.02 estimated.
1963:  2.10 actual, 2.06 estimated.
1964:  3.75 actual, 3.89 estimated.
1965:  3.81 actual, 3.77 estimated.
1966:  3.98 actual, 4.32 estimated.
1967:  4.39 actual, 4.81 estimated.
1968:  3.03 actual, 3.03 estimated.

Not too shabby, even if I say so myself.  

(A Note: In calculating ER% for a pitcher who played for two or more teams, one should try to calculate the pitcher ERs separately for each team he was on.  Many times this is not possible, so some sort of average should be used.)

A Modest Proposal

I propose that we use this method of calculating earned runs for pitchers, teams and leagues for which we don’t have ERAs in much the same manner that was used by the first edition of The Baseball Encyclopedia.  When the editors were not able to account for runs allowed by a pitcher as being either earned and unearned, they used italics.  Thus, when we use estimated figures for earned runs and earned run average, we should also use italicized numbers.

Additionally, I propose that if more that 10% of a pitcher’s years contain estimated ERAs, then his totals should also be italicized.

If we decide to employ this method of calculating ERAs, we will be able to fill in a great number of blanks in the playing records of some of the greatest pitchers who have ever played the game.  Moreover, this will— for the first time— allow us to compare pitchers from the the 19th Century, and the early part of this century, with pitchers who played the game in later years.

In conclusion, I propose that we use this method of calculating ERAs in the following instances:

  1. For pitchers in leagues where they did not calculate ERAs.

  2. For pitchers with less than 45 IP, where no ERAs were calculated by the league for less than’s.

  3. For team statistics, where the league did not calculate ERAs for those teams.

  4. For league statistics, where no ERAs were calculated by the league for the league in question.

I further propose that any future edition of The Minor League Register carry estimated ERAs; that members calculate and include estimated ERAs when disseminating Final League Averages that they have compiled from box scores.  (If we do use estimated ERAs, we would have to affix a note fully explaining what we are doing anytime we used them.)

In conclusion, I’d like to get some feedback from the membership on this, so here’s a ballot that I expect all members will send back.  We have important things to do on this committee, and some things that have to be decided— I’d like to see everybody on the Minor League Committee get their say, and have their ideas have some effect.  I think this is one way to do it.

Also, if any members have comments, suggestions or whatever, I’d like to especially hear from them on a separate sheet of paper.

The following was aproved by the minor league committee by a 35-3 vote.

Tommorrow I will write about what I have learned since this article was published.  Additionally, I will post a new chart that incorporates the research I have done since then.


Post a Comment

<< Home